Zurich and Cytora are using AI to turn commercial insurance submissions, often sent in different formats and languages, into structured information for underwriters. In the reported rollout, triage time dropped from 75 minutes to 15, digitization accuracy reached 98%, and straight-through intake processing rose from 10% to 95%. The AI helps organize submissions; underwriters still make the risk decisions.
The paperwork behind each submission
A commercial insurance submission might arrive as an email, a form, a spreadsheet, or a bundle of supporting documents. Underwriters have to pull out the relevant details, enter information into their systems, and work out which submissions need attention first.
It’s careful work, but it can take time away from assessing risks and talking with brokers. Zurich worked with Cytora to use AI for part of that intake process.
What the AI handles
Cytora’s platform reads incoming submissions and organizes the relevant details into structured information that fits Zurich’s underwriting approach. That information can then move into the underwriter’s workflow for review.
InsurTech Digital reported in May 2026 that Zurich deployed the platform in five countries over 90 days. The reported results were:
- Manual triage time fell from 75 minutes to 15 minutes per submission.
- Digitization accuracy rose from a reported 70–80% to 98%.
- Straight-through intake processing increased from 10% to 95%.
These figures were reported by the companies and industry coverage, rather than independently audited. And “straight-through” refers to processing submissions at intake. It doesn’t mean the AI makes the underwriting decision.
Why this can help
When submissions are organized sooner, underwriters can spend less time sorting paperwork and more time considering complex risks, working with brokers, and making decisions. Consistent data capture may also reduce duplicate entry and make it easier to prioritize submissions.
That’s the useful distinction here: AI prepares and routes information, while the underwriter remains responsible for assessing the risk.
How to put AI submission intake to work
For an insurer considering a similar system, it helps to start with one specific pain point. That could be the time spent processing submissions, repeated data entry, incomplete applications, or delays in getting a quote back to a broker.
From there, the work usually looks like this:
- Measure the current process. Record how long intake takes, how often data is re-entered, and where submissions tend to get stuck.
- Agree on the information underwriters need. Prepare representative documents and define the key fields. Leave room for differences between markets.
- Test extraction and validation. Compare AI-extracted details with human-reviewed submissions. Send uncertain or incomplete cases to staff.
- Connect the system to existing workflows. Route structured information into the tools underwriters already use, with a clear path for exceptions and human review.
- Start small, then assess the results. Pilot the workflow in one business segment or market. Check accuracy, time saved, exceptions, and whether staff find it useful before expanding.
A startup exploring this space could take the same measured approach with one specialty insurance line or broker channel. The first goal is to make one part of the process easier, not to automate every decision.
Where the opportunity is
There’s room for startups to make insurance information easier to handle: capture it once, organize it consistently, and get it to the right person sooner. That might mean helping insurers work with multilingual documents, different broker formats, or older underwriting systems.
Zurich’s rollout also points to the work behind a successful deployment. Clear data standards, adaptations for local markets, and input from underwriters all matter. The AI is one part of the system; fitting it into the way people actually work is what makes it useful.
You May Also Like
Aviva Puts Insurance Inside ChatGPT: How AI Is Changing Insurance Distribution in 2026
AI Transformation in Insurance Insurance acquisition is shifting from static web forms to conversational interfaces embedded inside daily AI tools. Aviva's deployment of a home and life insurance…
The STP Pivot: Why 2026 is the Year of the Autonomous Insurance Claim
For decades, "Insurance Automation" was a polite term for digital filing cabinets. In 2026, the industry has crossed a critical threshold where speed is no longer the goal,…
How Patra AI is Automating the Commercial Package Policies in 2026: A Guide for Insurance Teams Scaling to 85% Workflow Automation
How AI is finally automating complex commercial policies, modernizing workflows, and reinventing insurance operations for faster, smarter decisions In 2026, Patra AI’s breakthrough in insurance automation leapfrogged the…
AI in Insurance 2026: Rebuilding Trust Through Faster Claims and Smarter Fraud Detection
From near-instant claims approvals to fraud signals spotted before losses spiral, AI is steadily transforming insurance works in 2026. This shift is quiet. You will not find it…